Redirect Tests

A redirect test sends visitors assigned to a variant to a completely different URL. It’s the right pattern when the change you want to test is too large to express as inline CSS or JS — a full landing-page redesign, a different funnel entirely, a holiday-themed product page vs the regular one.

This guide covers when to use redirects, how Split Test Pro implements them, and the SEO discipline that keeps your search rankings safe.

When to Use a Redirect

Reach for a redirect when:

• You’re testing two distinct pages that share little markup. Editing one to look like the other via CSS/JS would be a thousand selectors.
• The variant is already built and live at its own URL — a new landing page, an alternate product detail page, an experimental funnel.
• You want to test the entire end-to-end experience, including JavaScript bundles and analytics setup that differ between the two pages.

Don’t use a redirect when:

• The change is small or visual. Use a CSS variant.
• The change involves logic on the same URL. Use a JS variant.
• You want to test only part of the page. Redirects move the visitor entirely.

How a Redirect Variant Works

Pick the variant on the Variants step and open its URL redirect tab. Under Destinations you get a card for the variant and one for the control, and the variant’s card has two modes:

• Full URL — type where visitors go under Visitors redirect to: a path on this site starting with / (for example /collections/all) or a full address starting with https:// for another site.
• Find & replace — type text under Find in the path and its replacement under Replace with. Visitors on a page whose path contains the text go to the same address with it replaced. It is plain text, not a pattern; only the first occurrence is replaced, and only in the path. As you type, the card shows a live example such as /summer-sale/shoes → /summer-sale-v2/shoes. Leave Replace with empty to remove the text.

Click Save on the card to store the redirect. The control’s card is marked Original: under Visitors stay on it shows the path of your Preview URL, locked, because control visitors are never redirected. The note under the cards confirms the scope — Only Variant B visitors are redirected, with and only on a page whose path matches added in Find & replace mode.

When a visitor is assigned to a redirect variant, the script applies the variant and the browser navigates to the new address. On a first visit the script first waits for the experiment configuration, so the visitor can see the original page briefly before the redirect; on repeat visits it redirects as soon as the script loads, from the configuration cached in the browser.

A Worked Example

You want to test your default product page (/products/widget) against a redesigned version (/products/widget-v2).

Targeting:

• Attribute: Path
• Operator: is
• Value: /products/widget

Variants — set Variant B up either way:

Variant	CSS	JS	Visitors redirect to
Control (A)	(empty)	(empty)	(stays on the page)
Variant B — Full URL	(empty)	(empty)	/products/widget-v2
Variant B — Find & replace	(empty)	(empty)	Find /widget, Replace with /widget-v2 → /products/widget-v2

When a visitor lands on /products/widget, half stay on the original; half are redirected to /products/widget-v2. With Find & replace, a visitor who arrived on /products/widget?utm_source=email keeps the utm_source parameter.

The SEO Trap (and How to Avoid It)

The biggest risk with redirect tests is search engines indexing your variant page as a duplicate of your control page, splitting your ranking signals or worse, ranking the variant.

Add noindex to the variant page

On the destination page (/products/widget-v2 in the example), add a robots meta tag:

<meta name="robots" content="noindex" />

This tells search engines not to index the variant. The visitors you redirect there won’t be affected — only crawlers will.

Use a canonical URL

If the variant page is structurally similar to the control, also set a canonical link to the control:

<link rel="canonical" href="https://example.com/products/widget" />

This tells search engines that the control is the authoritative version even if both pages get crawled.

Tracking Conversions on the Redirect Destination

For Shopify stores, the Web Pixel runs on every storefront page including redirected ones, so funnel events keep flowing. No extra setup.

For HTML sites, make sure the Split Test Pro snippet is also installed on the destination page — your conversion event needs to fire on the page where the conversion happens, not just on the source. If your destination page is a third-party domain you don’t control, redirect tests aren’t a great fit (you can’t track conversions there).

Things That Don’t Work With Redirects

• CSS or JS on the same variant. The redirect runs after the variant’s CSS and JavaScript, and then the visitor leaves the page. Switching the variant to the CSS or JavaScript tab and clicking Save removes the redirect: a variant is a redirect or code, not both.
• Redirect chains. Don’t redirect the destination page to yet another URL. The browser supports it but you’re adding latency for no reason and confusing analytics.
• Loops. If you target /a and redirect to /b, don’t also have an experiment that targets /b and redirects to /a. The visitor will bounce. If two running tests share visitors, the Overlap banner on Results shows it (Results Dashboard), but it’s worth being explicit: redirect tests are exclusive on a URL.
• Targeting that also matches the destination. The script runs again on the page you redirect to. With a Find & replace rule such as /widget → /widget-v2 and targeting Path contains /products/widget, the destination /products/widget-v2 matches too and is rewritten again. Target the source page exactly (Path is /products/widget).

Reading Results for a Redirect Test

Results work the same way as any other experiment — sessions and conversions are attributed to the variant the visitor was bucketed into, regardless of which URL they ended up on. A redirect variant’s visitors are the ones who landed on the source and were sent to the destination, not visitors who arrived at the destination directly.

Cleaning Up After a Winner

If your redirect variant wins:

1. Apply the destination page’s content to your default URL — i.e., make /products/widget look like /products/widget-v2 by editing your theme/site directly. Don’t keep the experiment running indefinitely just to maintain the redirect.
2. Either delete the source URL /products/widget-v2 or set up a permanent 301 server-side redirect for any backlinks that may have accumulated.
3. Remove the noindex from the now-default page so it’s indexable.

Next Steps

• For inline changes, see: CSS Variants and JavaScript Variants.
• Set up clean targeting that won’t accidentally fire your redirect on the destination page: URL Targeting.
• Avoid clobbering an existing test on the destination page: Running Multiple Experiments.